-----Original Message----- From: Ilpo Järvinen ij@kernel.org Sent: Thursday, June 12, 2025 11:17 PM To: Sasha Levin sashal@kernel.org; Chia-Yu Chang (Nokia) chia-yu.chang@nokia-bell-labs.com Cc: Eric Dumazet edumazet@google.com; stable@vger.kernel.org; stable-commits@vger.kernel.org; Neal Cardwell ncardwell@google.com; David S. Miller davem@davemloft.net; David Ahern dsahern@kernel.org; Jakub Kicinski kuba@kernel.org; Paolo Abeni pabeni@redhat.com; Simon Horman horms@kernel.org; Kuniyuki Iwashima kuniyu@google.com; Willem de Bruijn willemb@google.com Subject: Re: Patch "tcp: reorganize tcp_in_ack_event() and tcp_count_delivered()" has been added to the 6.6-stable tree
CAUTION: This is an external email. Please be very careful when clicking links or opening attachments. See the URL nok.it/ext for additional information.
- Chia-Yu
On Thu, 12 Jun 2025, Sasha Levin wrote:
On Thu, Jun 12, 2025 at 01:40:57AM -0700, Eric Dumazet wrote:
On Thu, May 22, 2025 at 3:44 PM Sasha Levin sashal@kernel.org wrote:
This is a note to let you know that I've just added the patch titled
tcp: reorganize tcp_in_ack_event() and tcp_count_delivered()to the 6.6-stable tree which can be found at: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fw ww.kernel.org%2Fgit%2F%3Fp%3Dlinux%2Fkernel%2Fgit%2Fstable%2Fstabl e-queue.git%3Ba%3Dsummary&data=05%7C02%7Cchia-yu.chang%40nokia-bel l-labs.com%7C449db2278c004aa84d7b08dda9f68c8c%7C5d4717519675428d91 7b70f44f9630b0%7C0%7C0%7C638853598557368335%7CUnknown%7CTWFpbGZsb3 d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFO IjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=Z6AfId4r6Ys1V4sGov 8bdOct72AAUdfVgFTo7NMOibU%3D&reserved=0
The filename of the patch is: tcp-reorganize-tcp_in_ack_event-and-tcp_count_delive.patch and it can be found in the queue-6.6 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree, please let stable@vger.kernel.org know about it.
May I ask why this patch was backported to stable versions ?
As you see Eric, you got no answer to a very direct question.
I've long since stopped caring unless a change really looks dangerous (this one didn't) what they take into stable, especially since they tend to ignore on-what-grounds questions.
This is causing a packetdrill test to fail.
Is this an issue upstream as well? Should we just drop it from stable?
It's long since I've done anything with packetdrill so it will take some time for me to test. Maybe Chia-Yu can check this faster (but I assume it's also problem in mainline as this is reported by Eric).
-- i.
Hi Eric,
I've checked the failure case and could reproduce it using the latest packetdrill.
The root cause is because delaying the tcp_in_ack_event() call does have an impact on update_alpha(), which uses the values of the latest delivered and delivered_ce updated by tcp_clean_rtx_queue(). Therefore, tcp_plb_update_state() will use these values to update the state for TCP PLB. While before this patch, update_alpha() is called before tcp_clean_rtx_queue(), and thus delivered and delivered_ce are not updated yet.
This is also in upstream as well. So, one question is why tcp_plb_update_state() uses non-latest delivered and delivered_ce before?
BRs, Chia-Yu